How to get the smallest value in submatrices

Hi,

I would appreciate if you could help me with this problem.

I have a big matrix , where are submatrices of the same size.

I want to get a vector containing the smallest elements in each submatrices. I want to avoid for loop as it might slow down my whole program.

Thanks.

Answer 1

[Bruno Luong]

B=randi(9,10,2)

B = 10×2

3 7 6 6 6 3 3 5 3 9 3 5 1 5 5 9 8 2 7 8

[m,n] = size(B);
k = 5;
if mod(m,k)
    error('m must divisible by k');
end
minsub = min(reshape(B,[m/k k n]),[],[1 3]).'

minsub = 5×1

3 3 3 1 2

[Hung Dao]

Thank you very much.

This is exactly what I need.

Answer 2

[Kevin Holly]

cellfun(@min,B,'UniformOutput',false)
%or   
   cellfun(@min,cellfun(@min,B,'UniformOutput',false),'UniformOutput',false)

[Hung Dao]

Thanks, Kevin

I have to put the bis matrix B into a cell array first, right?

[Kevin Holly]

Yes, so something like this:

B = {rand(1,4);rand(3,5);rand(4,5);rand(9,1);rand(2,4)}

B = 5×1 cell array
    {[0.5600 0.5471 0.0916 0.0913]}
    {3×5 double                   }
    {4×5 double                   }
    {9×1 double                   }
    {2×4 double                   }

cellfun(@min,B,'UniformOutput',false)

ans = 5×1 cell array
    {[                            0.0913]}
    {[0.5940 0.2929 0.2503 0.0334 0.1946]}
    {[0.2727 0.1998 0.0695 0.0921 0.1135]}
    {[                            0.0420]}
    {[       0.3392 0.4288 0.1739 0.4792]}

%or   
cellfun(@min,cellfun(@min,B,'UniformOutput',false),'UniformOutput',false)

ans = 5×1 cell array
    {[0.0913]}
    {[0.0334]}
    {[0.0695]}
    {[0.0420]}
    {[0.1739]}

This method allows you to find the minimum even if the submatrices where not the same size.

[Hung Dao]

Wonderful !

Thank you very much.

I use cell arrays but did not realise that it can be such useful.

[Hung Dao]

But, is there away to create this array without using for loop?

If I use a for loop to create this array then I could also just use forloop to find the minimum in each submatrices right?

[Kevin Holly]

I was under the impression that you already had a big matrix B that contained submatrices. If this is not the case, how are your data organized? Are all the matrices in a folder/workspace individually?

[Hung Dao]

My apologies for the confusion.

My algorithm leads to a big matrix, not the submatrices.

But I need to partition the big matrix into submatrices of the same size, and in each submatrix I want to find the minimum value.

I hope it is clear now.

Thank you.